www.nature.

com/scientificreports

OPEN The probabilistic pool punishment

proportional to the difference
of payoff outperforms previous
pool and peer punishment
Tetsushi Ohdaira
The public goods game is a multiplayer version of the prisoner’s dilemma game. In the public
goods game, punishment on defectors is necessary to encourage cooperation. There are two types
of punishment: peer punishment and pool punishment. Comparing pool punishment with peer
punishment, pool punishment is disadvantageous in comparison with peer punishment because pool
punishment incurs fixed costs especially if second-order free riders (those who invest in public goods
but do not punish defectors) are not punished. In order to eliminate such a flaw of pool punishment,
this study proposes the probabilistic pool punishment proportional to the difference of payoff. In the
proposed pool punishment, each punisher pays the cost to the punishment pool with the probability
proportional to the difference of payoff between his/her payoff and the average payoff of his/her
opponents. Comparing the proposed pool punishment with previous pool and peer punishment, in
pool punishment of previous studies, cooperators who do not punish defectors become dominant
instead of pool punishers with fixed costs. However, in the proposed pool punishment, more punishers
and less cooperators coexist, and such state is more robust against the invasion of defectors due to
mutation than those of previous pool and peer punishment. The average payoff is also comparable to
peer punishment of previous studies.

In the context of social dilemmas, it is widely known that a system of punishment is necessary for the evolution
of cooperation. Punishment is a factor that encourages cooperation, however, those who punish others lose
their fitness because the cost of punishment reduces their payoff. For this reason, whether punishment is really
necessary for the evolution of cooperation, especially whether a system of punishment will evolve, is highly
controversial. For example, some s­ tudies1–7 have a negative view of punishment, while o ­ thers8–18 have a positive
view of punishment.
In the previous studies that have shown a negative point of view, for example, in the prisoner’s dilemma game,
Dreber et al.2 state that the introduction of peer punishment results in the reduction in the average payoff of the
population as well as the introduction of pool punishment. Peer punishment is the one of two types of punish-
ment. Peer punishers directly punish uncooperative opponents by paying some cost. Pool punishment is the other
one. When pool punishers pay the investment, they also pay the cost to the punishment pool in advance. In the
public goods game, which is a multiplayer version of the prisoner’s dilemma game, Rand and N ­ owak5 extend the
public goods game to include the full set of punishment strategies and find that punishment no longer increases
cooperation, and that natural selection favours substantial levels of antisocial punishment for a wide range of
parameters. They also show that model predictions consistent with the results of behavioural experiments, and
punishment is mostly a self-interested tool for protecting oneself against potential competitors. N ­ owak6 states
that punishment is not a mechanism for the evolution of cooperation, but only complements other mechanisms
such as indirect reciprocity, group selection, and network reciprocity. Wu et al.7 describe that punishment does
not necessarily facilitate cooperation in one-to-one interactions such as the prisoner’s dilemma game.
In contrast to the preceding argument, regarding positive discussions, Traulsen et al.12 consider the public
goods game with cooperators, defectors, cooperative punishers, and the players who abstain from public goods
interactions. They show that cooperation (and punishment) is possible only if interactions are voluntary when
mutation rates are small. Sigmund et al.14 compare the prevailing model of peer punishment with pool punish-
ment. Pool punishment facilitates the sanction of second-order free-riders (who cooperate but do not sanction),

Institute of Information and Media, Aoyama Gakuin University, 5‑10‑1 Fuchinobe, Chuo‑ku, Sagamihara‑city,
Kanagawa 252‑5258, Japan. email: tetsushi77@gmail.com

Scientific Reports | (2022) 12:6604 | https://doi.org/10.1038/s41598-022-10582-5 1

Vol.:(0123456789)
 www.nature.com/scientificreports/

because those free-riders can be distinguished even if everyone contributes to the common good. Garcia and
­Traulsen16 state that Rand and N ­ owak5 discuss very limited cases where players refrain from collective action, and
show that cooperators who only punish defectors can thrive well when the abstainers are isolated even if players
have the choice of antisocial behaviour. Perc and S­ zolnoki17 propose the adaptive punishment that allows players
to change the degree to which they punish their opponents in proportion to the degree of success of cooperation.
The adaptive punishment promotes the reciprocity based on spatial connections between players (below simply
referred to as connections), and as a result, enhances cooperation. Nakamaru and D ­ ieckmann18 investigate the
evolution of social reaction norms and mainly show that the mechanism where evolution to enhanced coopera-
tion and stricter punishment reaction norms reinforce each other works best in the case of severe punishment.
The discussion regarding punishment is given in more detail below. Regarding peer punishment in the public
goods game, Helbing et al.19 show that the consideration of punishment enables us to understand the formation
and development of cooperators who punish defectors. Szolnoki et al.20 study the impact of pool punishment
on the spatial public goods game with cooperators, defectors, and pool punishers as three competitive strate-
gies. Helbing et al.19,21,22 specifically compare the efficiency of pool punishment with that of peer punishment
in maintaining social advantage. Chen et al.23 show that the introduction of punishment has a positive effect in
the public goods game, especially for large group-sized cooperation, but is not optimal for medium group-sized
cooperation. Sasaki et al.24 study deposits that will be refunded as long as committers adhere to the donation
game and punish free-riders and non-comitters.
We discuss other studies regarding punishment as follows. Gardner and W ­ est25 state that individuals show
greater cooperation when interacting with those who have the high possibility of punishing others. Egas and
­Riedl26 show that punishment is strongly dominated by its cost-to-impact ratio. Traulsen et al.27 present that
the majority of subjects would choose pool punishment if second-order free-riders would also be punished.
Schoenmakers et al.28 show that central agencies such as police can be crucial to the evolution of punishment.
­Perc29 shows that pool punishment in structured populations is sustainable, but it is limited to the case where
second-order free riders are also sanctioned to the extent that they cannot prevail. Chen and P ­ erc30 show that
the optimal distribution of resources within the framework of institutional punishment depends on whether
absolute or degree-normalized payoffs are used. Perc et al.31 systematically review the main results obtained in
the area of statistical physics of human cooperation and state that the problem of the cost of punishment can
be solved by probabilistically sharing responsibility for sanctioning defectors. In the spatial prisoner’s dilemma
game, ­Ohdaira32,33 shows that cooperation evolves not only in different types of spatial structures but also in the
case where both strategy and spatial structure evolve by changing the probability of punishment according to
the difference of payoff between players.
The alternative discussions regarding punishment are as follows. Szolnoki and P ­ erc34 consider traditional
cooperators and defectors, as well as cooperators punishing defectors and defectors punishing cooperators. They
show that antisocial punishment does not prevent cooperation if the synergistic effects are high enough to sustain
cooperation based on the network reciprocity and is viable only if the synergistic effects are low, punishment is
necessary for cooperation, and the cost-to-fine ratio is low. Chen and S­ zolnoki35 reveal that cooperators should
pay special attention to the growing capacity of renewable resources depending sensitively on the fraction of
cooperators and the total consumption of all players in addition to a delicately adjusted punishment. Lee et al.36
introduce a policelike or mercenary punisher who watches the population and punishes defectors and show that
the maximal average outcome can be reached at an intermediate cost value of punishment.
Here, this study especially focuses on peer punishment and pool punishment in the public goods game. As
described before, peer punishment means that a player pays the cost and directly imposes the punishment on a
defector, and pool punishment means that a player pays the cost to the punishment pool in advance. The advan-
tages and disadvantages of peer punishment and pool punishment are as follows. Peer punishment enables a
player to punish a defector directly, but it has the disadvantage of high cost of punishment. On the contrary, pool
punishment has a lower cost required for punishment than peer punishment, but it has the disadvantage that a
punisher must pay the cost to the punishment pool and fixed costs are incurred especially if we do not consider
the punishment on second-order free riders (those who invest in public goods but do not punish defectors).
In order to eliminate such a flaw of pool punishment, this study proposes the probabilistic pool punishment
proportional to the difference of payoff. In the proposed pool punishment, pool punisher compares his/her payoff
with the average payoff of the opponent public goods game participants and pays the cost to the punishment pool
only if he/she is disadvantageous in terms of payoff. This study considers the dynamics of four types of players,
i.e., punishers (contributing public goods and punishing defectors), defectors, cooperators (only contributing
public goods), and non-participants in the public goods game on the regular and the random one-dimensional
lattice with the average degree < k >  = 4. Then, we compare the proposed pool punishment with peer and pool
punishment of previous studies to investigate whether the flaw of pool punishment already mentioned can be
eliminated.

Model
The public goods game of this study is based on the framework by Traulsen et al.12 where investment in public
goods is distributed to all participants, including the invested player. When the number of participants in the
game is n, the number of cooperators is Nc, and the investment of cooperators in public goods is c, cooperators
will have the payoff of (rcNc/n)-c, while defectors will gain the payoff of rcNc/n. The value r is a factor multiply-
ing all summed-up contributions, and the best response of the participants is defection (not investing in public
goods). Therefore, in the public goods game, punishment on defectors is necessary to encourage cooperation. As
described before, there are two types of punishment: peer punishment and pool ­punishment14. Peer punishers
pay the investment c in public goods and then impose the punishment b on the defectors in the group by paying

Scientific Reports | (2022) 12:6604 | https://doi.org/10.1038/s41598-022-10582-5 2

Vol:.(1234567890)
www.nature.com/scientificreports/

Figure 1.  The sample spatial structure of (a) the regular and (b) the random one-dimensional lattice. The
number of players is 1000 in the actual simulations, but it is 20 in this figure. The meaning of each colour is as
follows: punishers are shown in blue, defectors are in red, cooperators are in green, and loners are in yellow.

the cost g. Therefore, when there are Ny defectors and Nw peer punishers in the group, defectors will be punished
with a sum of bNw, and peer punishers will bear the cost gNy. On the other hand, when pool punishers pay the
investment c, they also pay the cost G to the punishment pool in advance. Defectors will be fined the punishment
BNv proportional to the number of pool punishers, Nv.
Here, we consider the dynamics of four types of players, i.e., punishers (contributing public goods and pun-
ishing defectors), defectors, cooperators (only contributing public goods), and non-participants (below referred
to as loners) in the public goods game on the regular and the random one-dimensional lattice with the average
degree < k >  = 4. Figure 1a,b shows the sample spatial structure of the regular and the random one-dimensional
lattice. A vertex shows a player, and opponent players of each player in public goods game interactions are defined
by edges. Note that this figure has only 20 players so that we can easily grasp each spatial structure. The detail
of how to construct each spatial structure is described in the previous study of the a­ uthor32. To ensure that the
group of pool punishers or peer punishers outperforms the group of loners in terms of payoff, the payoff of loners
(σ) should be smaller than (r − 1)c-G in the case of pool punishment, or (r − 1)c in the case of peer punishment.
As described before, pool punishment is disadvantageous in comparison with peer punishment because pool
punishment incurs fixed costs especially if second-order free riders (those who invest in public goods but do not
punish defectors) are not punished. In order to eliminate such a flaw of pool punishment, this study proposes
the probabilistic pool punishment proportional to the difference of payoff. In the proposed pool punishment,
when the payoff of player i is Pi and the average payoff of the players with connections to player i is Pi , player i
pays the cost G to the punishment pool with the probability qi = (Pi − Pi ) / Pi ( Pi > 0). If Pi is equal to or larger
than 2Pi , then qi equals 1, and if Pi is smaller than Pi , then qi equals 0. Therefore, as shown in Fig. 2, if a player
has smaller payoff than the average payoff of opponent players, he/she contributes to the punishment pool with
high probability. On the other hand, if his/her payoff is nearly equal to the average payoff of opponent players, he/
she hardly contributes to the punishment pool. The previous ­study37 reports that the avoidance of overpunishing
(too much punishment on defectors with high payoff) is essential for the stable cooperation. In this study, to
avoid overpunishing, the payoff of each player will be 0 when it becomes a negative value.
The behaviour of player is updated according to the following rules. That is, each player i compares the new
payoff Pi′ with Pj′ of the players j in Oi after punishing the opponents and being punished by them. Note that Oi
  connected to player i. Then, each player i imitates
represents the set of all players   the behaviour of a player with
the highest payoff max Pj′ > Pi′ . If there are multiple players with max Pj′ , player i randomly imitates the
 
behaviour of the one of such
 players.
 If max Pj′ is equal toPi′ ,player i randomly switches his/her behaviour to
that of the player with max Pj′ including him/her. If max Pj′ is smaller than Pi′ , player i does not change his/
her behaviour. We consider a series of the public goods game, the process of imposing punishment, and the
imitation of behaviour as one generation. One simulation is executed up to 600 generations in order to reach a
sufficiently steady state, and we find the average value through 20 times simulations. Table 1 shows the specific
parameter settings required for simulations. The values of each parameter conform to Traulsen et al.12. Whenever
only a single cooperator or defector joins the game, he/she acts as a loner. That is, if only one group member
chooses to participate, then all group members receive the loner’s payoff σ. The value σ = 1 satisfies both condi-
tions σ < (r − 1)c-G in the case of pool punishment and σ < (r − 1)c in the case of peer punishment.

Scientific Reports | (2022) 12:6604 | https://doi.org/10.1038/s41598-022-10582-5 3

Vol.:(0123456789)
 www.nature.com/scientificreports/

If player i’s payoff is smaller

than the average of opponent
players’ payoff, player i pays
the cost to the punishment
pool with high probability.

Player i Opponent
players

If player i’s payoff is nearly

equal to the average of
opponent players’ payoff,
player i hardly pays the cost
to the punishment pool.
Figure 2.  This figure illustrates the characteristics of the probabilistic pool punishment proportional to the
difference of payoff.

A finite population (N) 1000

The average degree (number of connections) (< k >) 4
The average number of players participating in the public goods game (n) 5
A factor multiplying all summed-up contributions (r) 3
A cost to invest in a common good (c) 1
A reduction in the defector payoff for peer punishment (b) 1
Punishing a defector cost for peer punishment (g) 0.3
A reduction in the defector payoff for pool punishment (B) 1
Punishing a defector cost for pool punishment (G) 0.3
The loner payoff (σ) 1

Table 1.  Parameter settings required for simulations.

Results
Below, we compare the proposed pool punishment with previous pool and peer punishment. Firstly, Fig. 3a-c
shows the results of the regular one-dimensional lattice. Error bars indicate the standard deviation. (The follow-
ing figure also has error bars of SD). As those results show, in pool punishment of previous studies, cooperators
who do not punish defectors occupy almost the population instead of pool punishers who incur fixed costs. In
this case, the average payoff does not reach 3 because a few defectors remain. However, in the proposed pool
punishment, more punishers and less cooperators coexist. In peer punishment of previous studies, punishers
and cooperators coexist in almost the same number. Those results show that the proposed pool punishment is
more robust to the invasion of defectors due to mutation than peer punishment of previous studies. Besides,
in terms of the average payoff, the proposed pool punishment is almost the same as previous peer punishment.
Secondly, Fig. 4a-c shows the results of the random one-dimensional lattice. As those results show, in pool
punishment of previous studies, cooperators still dominate, although some punishers remain in the population
in comparison with the case of the regular one-dimensional lattice. On the other hand, in the proposed pool
punishment, because the payoff of each player is not averaged by his/her number of edges (degree), the differ-
ence of payoff between players depending on the degree of each player is larger than in the case of the regular
lattice. Therefore, compared to the regular case, the superiority of punishers over cooperators is reduced, and the
number of punishers is reduced. Nevertheless, punishers and cooperators coexist in almost the same number, and
the robustness against the invasion of defectors due to mutation is maintained because the number of punishers

Scientific Reports | (2022) 12:6604 | https://doi.org/10.1038/s41598-022-10582-5 4

Vol:.(1234567890)
www.nature.com/scientificreports/

(a)
1 3
0.9
2.5

Rate of Each Player

0.8

Average Payoff
0.7 2
0.6
0.5 1.5
0.4
0.3 1
0.2 0.5
0.1
0 0

120
160
200
240
280
320
360
400
440
480
520
560
600
40
80
Generation
Punishers Defectors Cooperators
Loners Average Payoff

(b) (c)
1 3 1 3
0.9 0.9
2.5 2.5
Rate of Each Player

Rate of Each Player

0.8 0.8

Average Payoff

Average Payoff
0.7 2 0.7 2
0.6 0.6
0.5 1.5 0.5 1.5
0.4 0.4
0.3 1 0.3 1
0.2 0.5 0.2 0.5
0.1 0.1
0 0 0 0
0

0
40
80
120
160
200
240
280
320
360
400
440
480
520
560
600

40
80
120
160
200
240
280
320
360
400
440
480
520
560
600
Generation Generation
Punishers Defectors Cooperators Punishers Defectors Cooperators
Loners Average Payoff Loners Average Payoff

Figure 3.  This figure shows the results regarding (a) the proposed pool punishment, (b) previous pool
punishment, and (c) previous peer punishment on the regular one-dimensional lattice. Parameter settings are as
follows; N = 1000, < k >  = 4, n = 5, r = 3, c = 1, b = 1, g = 0.3, B = 1, G = 0.3, and σ = 1.

is the largest among three types of punishment. Like the results of the regular lattice, the average payoff of the
proposed pool punishment is almost the same as that of previous peer punishment. In previous peer punishment,
because the cost of punishment is high, if a punisher has many opponent defectors, those defectors cannot be
punished, and then cooperators will have an advantage over such punisher.
The above results show that the proposed pool punishment solves the fixed cost problem of pool punishment
of previous studies, and can build a robust state against the invasion of defectors due to mutation. In addition, as
the results on the random lattice show, unlike peer punishment of previous studies, punishment is also available
due to the low cost of punishment even when a punisher has many opponent defectors.

Discussion
In the case of the scale-free one-dimensional lattice, which is not mentioned in the results of this study, the author
describes what the consequences of the proposed pool punishment and previous pool and peer punishment
will be. We compare both cases where the player with the most connections with opponents (the highest degree
player) becomes a punisher or a cooperator after the elimination of defectors by punishers. In the case where the
highest degree player becomes a punisher, he/she has an advantage over cooperators because he/she can obtain
enough payoff to offset the cost of punishment. On the other hand, when he/she becomes a cooperator, he/she has
an advantage over punishers because punishers cannot gain enough payoff to offset the cost of punishment. For
this reason, in the scale-free one-dimensional lattice, the number of simulations where punishers finally have an
advantage and such number where cooperators dominate at last is almost the same regarding the proposed pool
punishment and previous pool and peer punishment. This study does not consider punishment on cooperators,
the so-called second-order punishment. Therefore, if the number of connections with opponents is the same and
the number of punishers or cooperators in opponents is also the same, the highest degree cooperator always has
an advantage over punishers in terms of payoff. Considering the second-order punishment, the magnitude rela-
tion of payoff between cooperators and punishers naturally changes, then the result can be expected to change.
The author describes the difference between the proposed pool punishment and other probabilistic pun-
ishment as follows. Chen et al.38 consider probabilistic punishment as the simplest way of distributing the
responsibility to sanction defectors. The probability of punishment is fixed among players and does not change
depending on the difference of payoff like this study. The following studies also discuss probabilistic punishment:
class-specific probabilities of punishment that is based on the fixed number of c­ lasses39, the implicated punish-
ment that has a working probability p (0 < p < 1) and includes the peer punishment on defectors with a probability
q (0 < q < 1)40. However, those probabilities are fixed and also do not change. Szolnoki and ­Perc41 consider the
conditional punishment that does not depend on the difference of payoff like this study, but is proportional to

Scientific Reports | (2022) 12:6604 | https://doi.org/10.1038/s41598-022-10582-5 5

Vol.:(0123456789)
 www.nature.com/scientificreports/

(a)
1 3
0.9
2.5

Rate of Each Player

0.8

Average Payoff
0.7 2
0.6
0.5 1.5
0.4
0.3 1
0.2 0.5
0.1
0 0

120
160
200
240
280
320
360
400
440
480
520
560
600
40
80
Generation
Punishers Defectors Cooperators
Loners Average Payoff

(b) (c)
1 3 1 3
0.9 0.9
2.5 2.5
Rate of Each Player

Rate of Each Player

0.8 0.8

Average Payoff

Average Payoff
0.7 2 0.7 2
0.6 0.6
0.5 1.5 0.5 1.5
0.4 0.4
0.3 1 0.3 1
0.2 0.5 0.2 0.5
0.1 0.1
0 0 0 0
0

0
40
80
120
160
200
240
280
320
360
400
440
480
520
560
600

40
80
120
160
200
240
280
320
360
400
440
480
520
560
600
Generation Generation
Punishers Defectors Cooperators Punishers Defectors Cooperators
Loners Average Payoff Loners Average Payoff

Figure 4.  This figure shows the results regarding (a) the proposed pool punishment, (b) previous pool
punishment, and (c) previous peer punishment on the random one-dimensional lattice. Parameter settings are
as follows; N = 1000, < k >  = 4, n = 5, r = 3, c = 1, b = 1, g = 0.3, B = 1, G = 0.3, and σ = 1.

the number of other conditional and unconditional punishers within the group. The proposed pool punishment
is similar to Fehr and Schmidt’s inequity a­ version42 that players resist inequitable outcomes, i.e., they are willing
to give up their payoff to realize more equitable outcomes. However, in their study, a player can punish all other
players rather than other players having connections with him/her like this study.
Another similar method like the probability of punishment of this study is the emotional ­profile43,44. Szolnoki
et al.43 introduce sympathy and envy as the two emotional profiles that determine the strategy of each player,
and define them as the probability that each player cooperates with players having lower and higher payoff,
respectively. The evolutionary process leads to a spontaneous fixation to a single emotional profile; however,
this emotional profile depends not only on the payoff but also on the heterogeneity of connections between each
player. Szolnoki et al.44 also consider the imitation of emotional profiles of neighbour players instead of pure
strategy. The emotional profile of each player is determined by two pivotal (not continuous) factors only, namely
how each player behaves towards less and more successful neighbour players. On the other hand, the probability
of punishment of this study is continuous and based on the difference of payoff.
The following studies, although essentially different, utilize a method similar to the probabilistic pool punish-
ment of this study. Iwasa and L ­ ee45 introduce the graduated punishment that the degree of punishment gradually
changes based on the damage by selfish behaviour and show that the graduated punishment is the most effective
rule in the evolution of cooperation when the action of a player is incorrectly reported at a small probability and
the sensitivity of a player to the difference in the utility or payoff is not homogeneous. Helbing et al.21 investigate
the evolution of cooperation in the spatial public goods game and especially show that increasing the fine of
punishment induces a rising of the level of cooperation and larger punishment fines do not have any positive
effects. Jiang et al.46 also describe that severe punishment is not necessarily more effective and if cooperation is
likely, mild punishment leads to higher average payoffs.
This study proposes the probabilistic pool punishment proportional to the difference of payoff in order to
eliminate the flaw of pool punishment in which pool punishers incur fixed costs especially if second-order free
riders (those who invest in public goods but do not punish defectors) are not punished. In the proposed pool
punishment, each player pays the cost to the punishment pool with the probability depending on the differ-
ence between his/her payoff and the average payoff of the players with connections to him/her. Comparing the
proposed pool punishment with previous pool and peer punishment, in pool punishment of previous studies,
cooperators who do not punish defectors become dominant instead of pool punishers who incur fixed costs.
However, in the proposed pool punishment, more punishers and less cooperators coexist, and such state is more
robust against the invasion of defectors due to mutation than those of previous pool and peer punishment. The
average payoff is also comparable to peer punishment of previous studies.

Scientific Reports | (2022) 12:6604 | https://doi.org/10.1038/s41598-022-10582-5 6

Vol:.(1234567890)
www.nature.com/scientificreports/

In the future, the author will investigate whether the proposed pool punishment similarly does not allow the
invasion of defectors due to mutation and can maintain high average payoff in the cases where second-order
free riders are ­punished14, or all types of players can punish other ­players5. The author also intends to devise
the probabilistic pool reward and introduce the combination of reward and punishment like the following
previous ­studies47–49. Szolnoki and P­ erc47 discuss whether the combined application of reward and punishment
is evolutionary advantageous, and find rich dynamical behaviour that shows intricate phase diagrams where
continuous and discontinuous phase transitions successively occur. Chen et al.48 also propose the institutional
sanctioning policy that switches the incentive from rewarding to punishing when the frequency of cooperators
exceeds a threshold. They find that this policy establishes and recovers full cooperation at lower cost and under
a wider range of conditions than either rewards or penalties alone. Góis et al.49 show similar results that rewards
(positive incentives) are essential to initiate cooperation and sanctions (negative incentives) are instrumental to
maintain cooperation. As each parameter value of this study conforms to Traulsen et al.12, a factor multiplying
all summed-up contributions (r) equals 3, which is relatively large and somewhat induces cooperation. It is also
a future work to investigate whether the proposed pool punishment shows good results in the case of low r value
(e.g. r = 2) where cooperation does not easily evolve.

Received: 17 February 2022; Accepted: 8 April 2022

References
1. Sigmund, K., Hauert, C. & Nowak, M. A. Reward and punishment. Proc. Natl. Acad. Sci. U.S.A. 98, 10757–10762. https://​doi.​org/​
10.​1073/​pnas.​16115​5698 (2001).
2. Dreber, A., Rand, D. G., Fudenberg, D. & Nowak, M. A. Winners don’t punish. Nature 452, 348–351. https://d ​ oi.o
​ rg/1​ 0.1​ 038/n
​ atur​
e06723 (2008).
3. Herrmann, B., Thöni, C. & Gächter, S. Antisocial punishment across societies. Science 319, 1362–1367. https://​doi.​org/​10.​1126/​
scien​ce.​11538​08 (2008).
4. Ohtsuki, H., Iwasa, Y. & Nowak, M. A. Indirect reciprocity provides only a narrow margin of efficiency for costly punishment.
Nature 457, 79–82. https://​doi.​org/​10.​1038/​natur​e07601 (2009).
5. Rand, D. G. & Nowak, M. A. The evolution of antisocial punishment in optional public goods games. Nat. Commun. 2, 434. https://​
doi.​org/​10.​1038/​ncomm​s1442 (2011).
6. Nowak, M. A. Five rules for the evolution of cooperation. Science 314, 1560–1563. https://d ​ oi.o
​ rg/1​ 0.1​ 126/s​ cienc​ e.1​ 13375​ 5 (2006).
7. Wu, J.-J. et al. Costly punishment does not always increase cooperation. Proc. Natl. Acad. Sci. U.S.A. 106, 17448–17451. https://​
doi.​org/​10.​1073/​pnas.​09059​18106 (2009).
8. Fehr, E. & Gächter, S. Altruistic punishment in humans. Nature 415, 137–140. https://​doi.​org/​10.​1038/​41513​7a (2002).
9. Fehr, E. & Fischbacher, U. The nature of human altruism. Nature 425, 785–791. https://​doi.​org/​10.​1038/​natur​e02043 (2003).
10. Fowler, J. H. Altruistic punishment and the origin of cooperation. Proc. Natl. Acad. Sci. U.S.A. 102, 7047–7049. https://​doi.​org/​10.​
1073/​pnas.​05009​38102 (2005).
11. O’Gorman, R., Henrich, J. & Van Vugt, M. Constraining free riding in public goods games: Designated solitary punishers can
sustain human cooperation. Proc. R. Soc. B 276, 323–329. https://​doi.​org/​10.​1098/​rspb.​2008.​1082 (2009).
12. Traulsen, A., Hauert, C., De Silva, H., Nowak, M. A. & Sigmund, K. Exploration dynamics in evolutionary games. Proc. Natl. Acad.
Sci. U.S.A. 106, 709–712. https://​doi.​org/​10.​1073/​pnas.​08084​50106 (2009).
13. Rankin, D. J., Santos, M. D. & Wedekind, C. The evolutionary significance of costly punishment is still to be demonstrated. Proc.
Natl. Acad. Sci. U.S.A. 106, E135. https://​doi.​org/​10.​1073/​pnas.​09119​90107 (2009).
14. Sigmund, K., De Silva, H., Traulsen, A. & Hauert, C. Social learning promotes institutions for governing the commons. Nature
466, 861–863. https://​doi.​org/​10.​1038/​natur​e09203 (2010).
15. Szolnoki, A., Szabó, G. & Czakó, L. Competition of individual and institutional punishments in spatial public goods games. Phys.
Rev. E 84, 046106. https://​doi.​org/​10.​1103/​PhysR​evE.​84.​046106 (2011).
16. Garcia, J. & Traulsen, A. Leaving the loners alone: Evolution of cooperation in the presence of antisocial punishment. J. Theor. Biol.
307, 168–173. https://​doi.​org/​10.​1016/j.​jtbi.​2012.​05.​011 (2012).
17. Perc, M. & Szolnoki, A. Self-organization of punishment in structured populations. New J. Phys. 14, 043013. https://​doi.​org/​10.​
1088/​1367-​2630/​14/4/​043013 (2012).
18. Nakamaru, M. & Dieckmann, U. Runaway selection for cooperation and strict-and-severe punishment. J. Theor. Biol. 257, 1–8.
https://​doi.​org/​10.​1016/j.​jtbi.​2008.​09.​004 (2009).
19. Helbing, D., Szolnoki, A., Perc, M. & Szabó, G. Evolutionary establishment of moral and double moral standards through spatial
interactions. PLoS Comput. Biol. 6, e1000758. https://​doi.​org/​10.​1371/​journ​al.​pcbi.​10007​58 (2010).
20. Szolnoki, A., Szabó, G. & Perc, M. Phase diagrams for the spatial public goods game with pool punishment. Phys. Rev. E 83, 036101.
https://​doi.​org/​10.​1103/​PhysR​evE.​83.​036101 (2011).
21. Helbing, D., Szolnoki, A., Perc, M. & Szabó, G. Punish, but not too hard: How costly punishment spreads in the spatial public
goods game. New J. Phys. 12, 083005. https://​doi.​org/​10.​1088/​1367-​2630/​12/8/​083005 (2010).
22. Helbing, D., Szolnoki, A., Perc, M. & Szabó, G. Defector-accelerated cooperativeness and punishment in public goods games with
mutations. Phys. Rev. E 81, 057104. https://​doi.​org/​10.​1103/​PhysR​evE.​81.​057104 (2010).
23. Chen, X., Sasaki, T. & Perc, M. Evolution of public cooperation in a monitored society with implicated punishment and within-
group enforcement. Sci. Rep. 5, 17050. https://​doi.​org/​10.​1038/​srep1​7050 (2015).
24. Sasaki, T., Okada, I., Uchida, S. & Chen, X. Commitment to cooperation and peer punishment: Its evolution. Games 6, 574–587.
https://​doi.​org/​10.​3390/​g6040​574 (2015).
25. Gardner, A. & West, S. A. Cooperation and punishment, especially in humans. Am. Nat. 164, 753–764. https://​doi.​org/​10.​1086/​
425623 (2004).
26. Egas, M. & Riedl, A. The economics of altruistic punishment and the maintenance of cooperation. Proc. R. Soc. B 275, 871–878.
https://​doi.​org/​10.​1098/​rspb.​2007.​1558 (2008).
27. Traulsen, A., Röhl, T. & Milinski, M. An economic experiment reveals that humans prefer pool punishment to maintain the com-
mons. Proc. R. Soc. B 279, 3716–3721. https://​doi.​org/​10.​1098/​rspb.​2012.​0937 (2012).
28. Schoenmakers, S., Hilbe, C., Blasius, B. & Traulsen, A. Sanctions as honest signals—the evolution of pool punishment by public
sanctioning institutions. J. Theor. Biol. 356, 36–46. https://​doi.​org/​10.​1016/j.​jtbi.​2014.​04.​019 (2014).
29. Perc, M. Sustainable institutionalized punishment requires elimination of second-order free-riders. Sci. Rep. 2, 344. https://​doi.​
org/​10.​1038/​srep0​0344 (2012).
30. Chen, X. & Perc, M. Optimal distribution of incentives for public cooperation in heterogeneous interaction environments. Front.
Behav. Neurosci. 8, 248. https://​doi.​org/​10.​3389/​fnbeh.​2014.​00248 (2014).

Scientific Reports | (2022) 12:6604 | https://doi.org/10.1038/s41598-022-10582-5 7

Vol.:(0123456789)
 www.nature.com/scientificreports/

31. Perc, M. et al. Statistical physics of human cooperation. Phys. Rep. 687, 1–51. https://d ​ oi.o
​ rg/1​ 0.1​ 016/j.p
​ hysre​ p.2​ 017.0​ 5.0​ 04 (2017).
32. Ohdaira, T. Evolution of cooperation by the introduction of the probabilistic peer-punishment based on the difference of payoff.
Sci. Rep. 6, 25413. https://​doi.​org/​10.​1038/​srep2​5413 (2016).
33. Ohdaira, T. A remarkable effect of the combination of probabilistic peer-punishment and coevolutionary mechanism on the
evolution of cooperation. Sci. Rep. 7, 12448. https://​doi.​org/​10.​1038/​s41598-​017-​12742-4 (2017).
34. Szolnoki, A. & Perc, M. Second-order free-riding on antisocial punishment restores the effectiveness of prosocial punishment.
Phys. Rev. X 7, 041027. https://​doi.​org/​10.​1103/​PhysR​evX.7.​041027 (2017).
35. Chen, X. & Szolnoki, A. Punishment and inspection for governing the commons in a feedback-evolving game. PLoS Comput. Biol.
14, e1006347. https://​doi.​org/​10.​1371/​journ​al.​pcbi.​10063​47 (2018).
36. Lee, H.-W., Cleveland, C. & Szolnoki, A. Mercenary punishment in structured populations. Appl. Math. Comput. 417, 126797.
https://​doi.​org/​10.​1016/j.​amc.​2021.​126797 (2022).
37. Dercole, F., De Carli, M., Della Rossa, F. & Papadopoulos, A. V. Overpunishing is not necessary to fix cooperation in voluntary
public goods games. J. Theor. Biol. 326, 70–81. https://​doi.​org/​10.​1016/j.​jtbi.​2012.​11.​034 (2013).
38. Chen, X., Szolnoki, A. & Perc, M. Probabilistic sharing solves the problem of costly punishment. New J. Phys. 16, 083016. https://​
doi.​org/​10.​1088/​1367-​2630/​16/8/​083016 (2014).
39. Chen, X., Szolnoki, A. & Perc, M. Competition and cooperation among different punishing strategies in the spatial public goods
game. Phys. Rev. E 92, 012819. https://​doi.​org/​10.​1103/​PhysR​evE.​92.​012819 (2015).
40. Perc, M. & Szolnoki, A. A double-edged sword: Benefits and pitfalls of heterogeneous punishment in evolutionary inspection
games. Sci. Rep. 5, 11027. https://​doi.​org/​10.​1038/​srep1​1027 (2015).
41. Szolnoki, A. & Perc, M. Effectiveness of conditional punishment for the evolution of public cooperation. J. Theor. Biol. 325, 34–41.
https://​doi.​org/​10.​1016/j.​jtbi.​2013.​02.​008 (2013).
42. Fehr, E. & Schmidt, K. M. A theory of fairness, competition, and cooperation. Quart. J. Econ. 114, 817–868 (1999).
43. Szolnoki, A., Xie, N.-G., Ye, Y. & Perc, M. Evolution of emotions on networks leads to the evolution of cooperation in social dilem-
mas. Phys. Rev. E 87, 042805. https://​doi.​org/​10.​1103/​PhysR​evE.​87.​042805 (2013).
44. Szolnoki, A., Xie, N.-G., Wang, C. & Perc, M. Imitating emotions instead of strategies in spatial games elevates social welfare.
Europhys. Lett. 96, 38002. https://​doi.​org/​10.​1209/​0295-​5075/​96/​38002 (2011).
45. Iwasa, Y. & Lee, J.-H. Graduated punishment is efficient in resource management if people are heterogeneous. J. Theor. Biol. 333,
117–125. https://​doi.​org/​10.​1016/j.​jtbi.​2013.​05.​007 (2013).
46. Jiang, L.-L., Perc, M. & Szolnoki, A. If cooperation is likely punish mildly: Insights from economic experiments based on the
snowdrift game. PLoS ONE 8, e64677. https://​doi.​org/​10.​1371/​journ​al.​pone.​00646​77 (2013).
47. Szolnoki, A. & Perc, M. Correlation of positive and negative reciprocity fails to confer an evolutionary advantage: Phase transitions
to elementary strategies. Phys. Rev. X 3, 041021. https://​doi.​org/​10.​1103/​PhysR​evX.3.​041021 (2013).
48. Chen, X., Sasaki, T., Brännström, Å. & Dieckmann, U. First carrot, then stick: How the adaptive hybridization of incentives pro-
motes cooperation. J. R. Soc. Interface 12, 20140935. https://​doi.​org/​10.​1098/​rsif.​2014.​0935 (2014).
49. Góis, A. R., Santos, F. P., Pacheco, J. M. & Santos, F. C. Reward and punishment in climate change dilemmas. Sci. Rep. 9, 16193.
https://​doi.​org/​10.​1038/​s41598-​019-​52524-8 (2019).

Acknowledgements
This work was supported by JSPS KAKENHI Grant Numbers JP17K18074 and JP20K11959. The author truly
thanks two anonymous referees for their positive comments and kind suggestions.

Author contributions
T.O. designed and performed the research, analysed the data and wrote the paper.

Competing interests
The author declares no competing interests.

Additional information
Correspondence and requests for materials should be addressed to T.O.
Reprints and permissions information is available at www.nature.com/reprints.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International
License, which permits use, sharing, adaptation, distribution and reproduction in any medium or
format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the
Creative Commons licence, and indicate if changes were made. The images or other third party material in this
article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the
material. If material is not included in the article’s Creative Commons licence and your intended use is not
permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder. To view a copy of this licence, visit http://​creat​iveco​mmons.​org/​licen​ses/​by/4.​0/.

© The Author(s) 2022

Scientific Reports | (2022) 12:6604 | https://doi.org/10.1038/s41598-022-10582-5 8

Vol:.(1234567890)